@Test
public void testVCFHeaderSerialization() throws Exception {
final VCFFileReader reader =
new VCFFileReader(new File("testdata/htsjdk/variant/HiSeq.10000.vcf"), false);
final VCFHeader originalHeader = reader.getFileHeader();
reader.close();
final VCFHeader deserializedHeader = TestUtil.serializeAndDeserialize(originalHeader);
Assert.assertEquals(
deserializedHeader.getMetaDataInInputOrder(),
originalHeader.getMetaDataInInputOrder(),
"Header metadata does not match before/after serialization");
Assert.assertEquals(
deserializedHeader.getContigLines(),
originalHeader.getContigLines(),
"Contig header lines do not match before/after serialization");
Assert.assertEquals(
deserializedHeader.getFilterLines(),
originalHeader.getFilterLines(),
"Filter header lines do not match before/after serialization");
Assert.assertEquals(
deserializedHeader.getFormatHeaderLines(),
originalHeader.getFormatHeaderLines(),
"Format header lines do not match before/after serialization");
Assert.assertEquals(
deserializedHeader.getIDHeaderLines(),
originalHeader.getIDHeaderLines(),
"ID header lines do not match before/after serialization");
Assert.assertEquals(
deserializedHeader.getInfoHeaderLines(),
originalHeader.getInfoHeaderLines(),
"Info header lines do not match before/after serialization");
Assert.assertEquals(
deserializedHeader.getOtherHeaderLines(),
originalHeader.getOtherHeaderLines(),
"Other header lines do not match before/after serialization");
Assert.assertEquals(
deserializedHeader.getGenotypeSamples(),
originalHeader.getGenotypeSamples(),
"Genotype samples not the same before/after serialization");
Assert.assertEquals(
deserializedHeader.samplesWereAlreadySorted(),
originalHeader.samplesWereAlreadySorted(),
"Sortedness of samples not the same before/after serialization");
Assert.assertEquals(
deserializedHeader.getSampleNamesInOrder(),
originalHeader.getSampleNamesInOrder(),
"Sorted list of sample names in header not the same before/after serialization");
Assert.assertEquals(
deserializedHeader.getSampleNameToOffset(),
originalHeader.getSampleNameToOffset(),
"Sample name to offset map not the same before/after serialization");
Assert.assertEquals(
deserializedHeader.toString(),
originalHeader.toString(),
"String representation of header not the same before/after serialization");
}
@DataProvider(name = "HiSeqVCFHeaderDataProvider")
public Object[][] getHiSeqVCFHeaderData() {
final File vcf = new File("testdata/htsjdk/variant/HiSeq.10000.vcf");
final VCFFileReader reader = new VCFFileReader(vcf, false);
final VCFHeader header = reader.getFileHeader();
reader.close();
return new Object[][] {{header}};
}
@Test
public void shouldPreserveSymbolicAlleleCase() {
VCFFileReader reader =
new VCFFileReader(new File(VariantBaseTest.variantTestDataRoot + "breakpoint.vcf"), false);
VariantContext variant = reader.iterator().next();
reader.close();
// VCF v4.1 s1.4.5
// Tools processing VCF files are not required to preserve case in the allele String, except for
// IDs, which are case sensitive.
Assert.assertTrue(variant.getAlternateAllele(0).getDisplayString().contains("chr12"));
}
@Test
public void testVCFHeaderAddDuplicateContigLine() {
File input = new File("testdata/htsjdk/variant/ex2.vcf");
VCFFileReader reader = new VCFFileReader(input, false);
VCFHeader header = reader.getFileHeader();
final int numContigLinesBefore = header.getContigLines().size();
// try to readd the first contig line
header.addMetaDataLine(header.getContigLines().get(0));
final int numContigLinesAfter = header.getContigLines().size();
// assert that we have the same number of contig lines before and after
Assert.assertEquals(numContigLinesBefore, numContigLinesAfter);
}
@Test
public void testVCFHeaderAddDuplicateHeaderLine() {
File input = new File("testdata/htsjdk/variant/ex2.vcf");
VCFFileReader reader = new VCFFileReader(input, false);
VCFHeader header = reader.getFileHeader();
VCFHeaderLine newHeaderLine = new VCFHeaderLine("key", "value");
// add this new header line
header.addMetaDataLine(newHeaderLine);
final int numHeaderLinesBefore = header.getOtherHeaderLines().size();
// readd the same header line
header.addMetaDataLine(newHeaderLine);
final int numHeaderLinesAfter = header.getOtherHeaderLines().size();
// assert that we have the same number of other header lines before and after
Assert.assertEquals(numHeaderLinesBefore, numHeaderLinesAfter);
}
@Test
public void testVCFHeaderAddMetaDataLineDoesNotDuplicateContigs() {
File input = new File("testdata/htsjdk/variant/ex2.vcf");
VCFFileReader reader = new VCFFileReader(input, false);
VCFHeader header = reader.getFileHeader();
final int numContigLinesBefore = header.getContigLines().size();
VCFInfoHeaderLine newInfoField =
new VCFInfoHeaderLine(
"test", VCFHeaderLineCount.UNBOUNDED, VCFHeaderLineType.String, "test info field");
header.addMetaDataLine(newInfoField);
// getting the sequence dictionary was failing due to duplicating contigs in issue #214,
// we expect this to not throw an exception
header.getSequenceDictionary();
final int numContigLinesAfter = header.getContigLines().size();
// assert that we have the same number of contig lines before and after
Assert.assertEquals(numContigLinesBefore, numContigLinesAfter);
}